An extremely large cube of data may be distributed across a multi-node grid of connected computing devices to reduce access times and to scale the storage space. Multiple users may access the cube of data at any given time. It may be impractical to store multiple copies of the cube of data due to the extremely large amount of disk space needed as well as the memory needed to load the cube of data for use by users. Memory mapping technology may be used so that, as multiple users access the same cube, the cube is loaded into memory only once.